Sound Semiconductor SSI2161

[neil.johnson]

Hi all,

During my travels through this magical life I have developed some involvement with the fine folks at Sound Semiconductor (www.soundsemiconductor.com).  They recently launched the SSI2161 single-channel VCA, as well as the SSI2162 dual VCA, both derived from the SSI2164 quad VCA, itself a modern take on the classic SSM2164.

The SSI2161 is a clear competitor to the THAT 2180, although slightly different control interface, and with variable core bias so you can dial in which kind of badness you can live with (less noise or less distortion) - or futz around with sliding bias if that's your thing.

While I am coming at this from the world of music synthesizers, I'm curious to hear folks' views on VCAs in modern recording/audio systems?  Compressors/limiters/expanders is the obvious application, is VCA automation still a thing these days?

Cheers,
Neil

 

[Bo Deadly]

Nice specs. Just a guess but it looks like SSI2161 is actually 4 VCAs in parallel, SSI2162 is 2 x 2 in parallel for 6dB and 3dB noise improvement respectively. Unfortunately it doesn't look like you can actually buy SSI2161 anywhere so it doesn't much matter.

 

[abbey road d enfer]

The SSI2161 is a clear competitor to the THAT 2180, although slightly different control interface, and with variable core bias so you can dial in which kind of badness you can live with (less noise or less distortion) - or futz around with sliding bias if that's your thing.

On paper it looks like they could perform quite similarly, except when I see the THD figures. At +15dBu output, teh 2161 is at 0.17% when the 2181B (the best price/performance compromise) is at less than 0.1%; this is in conditions that yield comparable noise performance.
There are other aspects that cannot be evaluated on paper, such as noise modulation and control feedthrough. It would take A/B testing.

While I am coming at this from the world of music synthesizers, I'm curious to hear folks' views on VCAs in modern recording/audio systems?  Compressors/limiters/expanders is the obvious application, is VCA automation still a thing these days?

Quite anecdotical, really. VCA automation makes sense in a full-analog recording chain, and storing the automation info remains the major issue. Since many people mix their analog tapes into a DAW, they use the DAW's facilities for automation.
However, many still use their VCA groups.

 

[neil.johnson]

Nice specs. Just a guess but it looks like SSI2161 is actually 4 VCAs in parallel, SSI2162 is 2 x 2 in parallel for 6dB and 3dB noise improvement respectively. Unfortunately it doesn't look like you can actually buy SSI2161 anywhere so it doesn't much matter.

OEMs/volume customers can buy direct from SSI.  For DIY small orders some of the SSI distributors have stock.

Neil

 

[Rob Flinn]

It appears it's only available as a 10 pin dual in line surface mount package from the data sheet, which if this is the case isn't ideal for DIY purposes, but not a deal breaker.

On paper it looks like they could perform quite similarly, except when I see the THD figures. At +15dBu output, teh 2161 is at 0.17% when the 2181B (the best price/performance compromise) is at less than 0.1%; this is in conditions that yield comparable noise performance.

I've been told in the past that it is better to only use VCA's in attenution because they have worse distortion when you ask them to provide gain.  Not sure how relevant that is with the most modern devices.

 

[abbey road d enfer]

I've been told in the past that it is better to only use VCA's in attenution because they have worse distortion when you ask them to provide gain.  Not sure how relevant that is with the most modern devices.

Gain is not the issue, level is.
Most VCA's have distortion characteristics that depends on the sum of input and output currents. In other words there is as much distortion due to 0dBu input and 20dB attenuation than -20dBu input and +20dB gain.
Indeed, the higher the level, the higher the distortion.

 

[zamproject]

Hello

It appears it's only available as a 10 pin dual in line surface mount package from the data sheet, which if this is the case isn't ideal for DIY purposes, but not a deal breaker.

Honestly I'm faster soldering SOIC package(1.27)  than TH...
I think it's a non issue for DIY

Best
Zam

 

[Rob Flinn]

Hello

Honestly I'm faster soldering SOIC package(1.27)  than TH...
I think it's a non issue for DIY

Best
Zam

Maybe you are but a lot of people here are beginners, so maybe more of an issue for them than a seasoned pro like yourself.

 

[JohnRoberts]

I've been told in the past that it is better to only use VCA's in attenution because they have worse distortion when you ask them to provide gain.  Not sure how relevant that is with the most modern devices.

As Abbey noted there are multiple moving parts... level in VCAs is typically currents and trying to scale current for lower distortion can deteriorate S/N ratio (no free lunch).

[TMI] Over a decade ago I speculated about a low distortion subtractive VCA topology for modest amounts of gain reduction.

Imagine a dry audio path for unity gain where the audio does not even touch the VCA (you can't do better than that for low noise and low distortion.  To command attenuation an inverted polarity audio signal coming from the VCA is summed with the dry path.

For -3dB attenuation the VCA path is feeding roughly -10dB polarity inverted level signal. For -6dB attenuation the VCA is also operating at -6dB. Deeper attenuation requires closer to unity gain VCA path for diminishing benefit.

This is not superior to a simple VCA for deep attenuation like fader automation, a standard VCA attenuates its own noise when commanding deep cut. OTOH for modest amounts of attenuation (like in a comp/limiter) this can be cleaner than a conventional VCA path.

I never melted solder over this but shared it with a good friend who was still designing consoles at the time (now RIP). He actually designed a nice limiter around this subtractive topology. [/TMI]

JR

PS: SMD is getting harder to ignore if you want to use the latest SOTA ICs. I use parts in my drum tuner that are not sold in TH versions.

 

[Bo Deadly]

Imagine a dry audio path for unity gain where the audio does not even touch the VCA (you can't do better than that for low noise and low distortion.  To command attenuation an inverted polarity audio signal coming from the VCA is summed with the dry path.

Interesting!

So is it an inverting op amp into second inverting op amp into input of VCA into - input of first op amp?

This is not superior to a simple VCA for deep attenuation like fader automation, a standard VCA attenuates its own noise when commanding deep cut. OTOH for modest amounts of attenuation (like in a comp/limiter) this can be cleaner than a conventional VCA path.

 

[JohnRoberts]

Interesting!

So is it an inverting op amp into second inverting op amp into input of VCA into - input of first op amp?

I am not sure what you mean... what I am trying to describe is two parallel paths,

the first dry unity gain non inverted path, and a second inverted (opposite polarity) path running through the VCA. The output of these two paths get summed together so the opposite polarity content reduces the dry normal content.

The benefit is that the VCA is mostly out of the signal path except for deep attenuation.

I hope that helps.

JR

 

[Bo Deadly]

Ah, ok. Understood.

This is what I was thinking:

 

[neil.johnson]

I am not sure what you mean... what I am trying to describe is two parallel paths,

the first dry unity gain non inverted path, and a second inverted (opposite polarity) path running through the VCA. The output of these two paths get summed together so the opposite polarity content reduces the dry normal content.

The benefit is that the VCA is mostly out of the signal path except for deep attenuation.

I hope that helps.

JR

Doug Self covered something that sounds like this in his Small Signal Audio book (2nd ed, chapter 19, figure 19.15).  That circuit would work with a THAT2180 as it inverts the current phase so you get the inversion for free, so really all you need to add to the standard THAT2180 circuit is a parallel resistor and then cope with the overall signal inversion.  Apparently the idea went into a broadcast console in 1990.

Neil

 

[JohnRoberts]

Doug Self covered something that sounds like this in his Small Signal Audio book (2nd ed, chapter 19, figure 19.15).  That circuit would work with a THAT2180 as it inverts the current phase so you get the inversion for free, so really all you need to add to the standard THAT2180 circuit is a parallel resistor and then cope with the overall signal inversion.  Apparently the idea went into a broadcast console in 1990.

Neil

The ancients keep stealing our ideas... 

JR

 

[neil.johnson]

The ancients keep stealing our ideas... 

JR

LOL!  I gather though that that circuit can suffer from offness problems as you need the direct and VCA paths to exactly cancel, which is tricky across component tolerance, thermal shifts, delays through the VCA, etc.

Neil

 

[abbey road d enfer]

LOL!  I gather though that that circuit can suffer from offness problems as you need the direct and VCA paths to exactly cancel, which is tricky across component tolerance, thermal shifts, delays through the VCA, etc.

JR mentioned that this scheme is adequate for moderate attenuation only, and not expect full attenuation.
Crown used this in their Microtech/Macrotech amplifiers for clip prevention. A simple OTA connected in the feedback. The OTA was triggered by the sudden voltage increase at the output of an opamp that happened when the amp ran out of steam, for whatever reason.

 

[JohnRoberts]

JR mentioned that this scheme is adequate for moderate attenuation only, and not expect full attenuation.

Yup, deep attenuation is difficult and actually offers no benefit because you have near 0dB VCA self noise... A normal VCA attenuates its own noise when commanding deep cut.

Crown used this in their Microtech/Macrotech amplifiers for clip prevention. A simple OTA connected in the feedback. The OTA was triggered by the sudden voltage increase at the output of an opamp that happened when the amp ran out of steam, for whatever reason.

That is different topology (I think?) where the OTA is parallel with the feedback resistor to reduce gain.  That said it does enjoy the benefit of being out of the path when cut off. For clip limiting it is hard to complain about sonic quality of limiting vs. clipping. I am not familiar with micro/macro tech schematics so will take your word for it. I actually designed a quad noise gate/limiter using that topology back in the 1980s with remarkable specs dry, and adequate when limiting/gating.

I am familiar with Peavey's DDT (clip limiting) that uses an OTA opposing a signal through an input resistance. The Peavey clip detector looked at the power amp overall negative feedback and sensed when the - input was no longer tracking the + input. This delivered a very sensitive clip detector, not to mention detecting current limiting or any other deviation of output from what it should be.

JR

 

[Newmarket]

Maybe you are but a lot of people here are beginners, so maybe more of an issue for them than a seasoned pro like yourself.

tbh once you get a bit used to SMT and the right basic kit - fine point iron / tweezers / magnification  - then it can be easier.
Finer pitch than 1,27mm can be a challenge although I've done t ( get some Solder Wick !) but you avoid the wear and tear of thru holes - expecting that rework is inherent in DIY - and it can be easier if only having components on one side of pcb.
Many new devices only available in SMT do need to get used to it,